48 REPORT— 1903, 



Thernldmetel' BAo, which was heated about fifty times during No veill* 

 bar 1902 in electric furnaces up to 1050°, and again during April and 

 May 1903 to similar temperatures for prolonged periods, appears to be 

 hardly perceptibly affected by it, no certa,in change of FI occurring during 

 the period February 12 to August 18 covered by the later experiments, 

 and certainly no variation of the zero of "1° C. 



To see if the small lack of homogeneity of the wire as shown by the 

 properties of the different thermometers was due to the treatment it had 

 received during the successive adjustments of FI, a new thermometer, 

 named BA7, was made up of wire taken from the inner end of the same 

 reel as the other six. No attempt was made at adjustment of its FI, 

 which was found after thorough annealing to be 100022 box units. 



The 8 was found to be 1'506, an intermediate value. The wire was 

 then unwound from the mica frame and suspended freely in air between 

 the ends of the leads, and a current of 2h amperes, whicli was sufficient to 

 maintain it at about 1400° C, was passed for about 2 hours. 



Owing to the volatilisation of a considerable quantity of platinum 

 from the wire, a large increase in the FI was found, as was expected, but 



the 8 remained unchanged, though a rise in -L was recorded amounting to 



1 part in 1000. 



In order to make certain that the differences observed were not due 

 to defective insulation in the thermometers, the insulation resistance be- 

 tween the thermometer and compensator leads of each of the thermometers 

 was measured by a direct deflection method, and found to be in no case 

 less than 700,000 ohms at any temperature between 0° and 1000° for BAj 

 and BA.2, and 0° and 500° for BA3 and BA^. Some experiments were 

 also made on an imitation platinum thermometer having its coil wound 

 on mica of standard quality, but cut at the lower end into two parts. 



Although the insulation from one part to another was practically infi- 

 nite at all temperatures, when only platinum and mica were present in 

 the heated part of the porcelain tube, the introduction of a small piece of 

 clean copper wire into the hot space near the bulb was sufficient after 

 some time to lower the insulation, even at only about 800° C, to a few 

 thousand ohms. The cause of the differences between the individual 

 thermometers does not, therefore, appear to be leakage. 



Neither does the cause of the small differences in values of 8 found lie 

 in the method of taking the sulphur point, as the same apparatus was 

 used in the same way for all the experiments. The sulphur is now boiled 

 in an ai'rangement similar to Callendar and Grifhths's well-known pattern, 

 except that, to avoid the necessity of removing the tube at each reheat 

 after the sulphur has crystallised, the glass boiling-tube is replaced by 

 one of thin weldless steel, brazed with spelter into a rather wider end- 

 piece of thick iron tubing, which is exposed to the direct flame of the 

 large bunsen used for heating. The level of the liquid sulphur is always 

 maintained at least 2 inches above the bottom plate of the apparatus, 

 and the upper level of the vapour to a definite position, which can be seen 

 through mica windows in the upper part of the neck. Under these con- 

 ditions no measurable superheating of the vapour has ever been observed, 

 and a comparison of the sulphur points obtained with this form of ap- 

 paratus with those got in the older one, with glass boiling-tube, reveals no 

 measurable systematic difference. 



For the boiling-point of sulphur under normal pressure in latitude 45' 



